We propose in this paper a novel subspace identification method, based on PARSIMonious parameterization (Qin et al., 2005), and we show that such algorithm guarantees consistent estimates of the Markov parameters with open-loop and closed-loop data. The method uses the predictor form and it effectively exploits in all steps the Toeplitz structure of the Markov parameters' matrices. After evaluation of (AK = A-KC,C) from the identified observability matrix, the method computes (BK = B-KD, D, K) and the initial condition by solving a single (well conditioned even for unstable systems) Least Squares problem. We use such method to obtain linear models for MPC design, and we show how the proposed method compares favorably with other existing subspace algorithms in two examples. © 2009 IFAC.
Closed-Loop PARSIMonious Subspace Identification: Theory and Application to MPC
PANNOCCHIA, GABRIELE;
2010-01-01
Abstract
We propose in this paper a novel subspace identification method, based on PARSIMonious parameterization (Qin et al., 2005), and we show that such algorithm guarantees consistent estimates of the Markov parameters with open-loop and closed-loop data. The method uses the predictor form and it effectively exploits in all steps the Toeplitz structure of the Markov parameters' matrices. After evaluation of (AK = A-KC,C) from the identified observability matrix, the method computes (BK = B-KD, D, K) and the initial condition by solving a single (well conditioned even for unstable systems) Least Squares problem. We use such method to obtain linear models for MPC design, and we show how the proposed method compares favorably with other existing subspace algorithms in two examples. © 2009 IFAC.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.